haber
Gyrfalcon launches AI accelerator with eMRAM.
Gyrfalcon Technology, the 11-month-old Milpitas specialist in AI accelerator ICs, has announced commercial availability of a 22nm ASIC with embedded MRAM called the LightspeeurÒ 2802M.
Gyrfalcon claims the chip is the industry’s first AI Accelerator to use MRAM.
The chip has a density of 50 F and and a read speed of 45-60ns.
It has 40MB of memory, which can support large AI models or multiple AI models within a single chip including image classification, voice identification, voice commands, facial recognition, pattern recognition and others.
“Our proprietary and patented MRAM engine has come to fruition in the LightspeeurÒ 2802M, the first of many MRAM–enabled AI accelerators that GTI plans to bring to market,” says GTI co-founder Terry Torng.
The LightspeeurÒ 2802M is the first of a portfolio of MRAM chips in the company’s roadmap.
The company plans to work on large–scale opportunities with customers to develop specific “super AI Accelerator chips” that combine the new MRAM Engine with GTI’s other patented innovations such as Matrix Processing Engine (MPE) and AI Processing in Memory (APiM) technologies.
The MPE utilizes CNN with the APiM, pairing logic with memory like a real human brain. This optimizes the speed of processing, achieving high TOPS performance, while saving power by avoiding management of data in discrete memory components.
The MRAM line of products will leverage all of the GTI tools and technical documentation created for other AI Accelerator chips made available earlier this year, such as the SDK and the recently announced PLAI accessories, which simplifies AI model creation for development teams less familiar with developing solutions using artificial intelligence.
AI models created with Caffe, TensorFlow and other tools can be supported, enabling a wide variety of neural networks, including VGG, MobileNet, ResNet, SDD, ImageNet and others.
Gyrfalcon claims the chip is the industry’s first AI Accelerator to use MRAM.
The chip has a density of 50 F and and a read speed of 45-60ns.
It has 40MB of memory, which can support large AI models or multiple AI models within a single chip including image classification, voice identification, voice commands, facial recognition, pattern recognition and others.
“Our proprietary and patented MRAM engine has come to fruition in the LightspeeurÒ 2802M, the first of many MRAM–enabled AI accelerators that GTI plans to bring to market,” says GTI co-founder Terry Torng.
The LightspeeurÒ 2802M is the first of a portfolio of MRAM chips in the company’s roadmap.
The company plans to work on large–scale opportunities with customers to develop specific “super AI Accelerator chips” that combine the new MRAM Engine with GTI’s other patented innovations such as Matrix Processing Engine (MPE) and AI Processing in Memory (APiM) technologies.
The MPE utilizes CNN with the APiM, pairing logic with memory like a real human brain. This optimizes the speed of processing, achieving high TOPS performance, while saving power by avoiding management of data in discrete memory components.
The MRAM line of products will leverage all of the GTI tools and technical documentation created for other AI Accelerator chips made available earlier this year, such as the SDK and the recently announced PLAI accessories, which simplifies AI model creation for development teams less familiar with developing solutions using artificial intelligence.
AI models created with Caffe, TensorFlow and other tools can be supported, enabling a wide variety of neural networks, including VGG, MobileNet, ResNet, SDD, ImageNet and others.